Noise reduction for low-dose helical CT by 3D penalized weighted least-squares sinogram smoothing

Wang, Jing; Li, Tianfang; Lu, Hongbing; Zhang, Liang

doi:10.1117/12.653844

Cited by 3 publications

(4 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The KL transform can manipulate a sequence of somewhat correlated measurements into an ordered series of principal components and provide a unique means for noise reduction, feature extraction and de-correlation [6]. The Karhunen-Loève (KL) transform is used to decompose measured signals into uncorrelated empirical basis functions.…”

Section: Karhunen-loève Transformmentioning

confidence: 99%

“…When the co-occurrence matrix is obtained, the process of kl transform can be implemented by computing the covariance matrix Ci for each column in the cooccurrence matrix, see equation (9)[6], using the mean vector μ that calculated using equation ( 8) [6], Where: Xi is the ith column in the co-occurrence matrix, is the mean vector, N is the number of columns.…”

Section: Kl-transformmentioning

confidence: 99%

“…The target of proposed neural network is either +1 indicating to the existence of hidden information in the image, or -1 to represent that the image is clean. Figure (6) shows the architecture of network used to implement the proposed system.…”

Section: Applying Rbnn To Detect Information Hidingmentioning

confidence: 99%

See 2 more Smart Citations

Steganalysis Using KL Transform and Radial Basis Neural Network

Hasoon¹,

Khalifa²

2012

AL-Rafidain Journal of Computer Sciences and Mathematics

View full text Add to dashboard Cite

The essential problem in the security field is how to detect information hiding. This paper proposes a new steganalysis scheme based on artificial neural network as a classifier to detect information hiding in colored and grayscale images. The statistical features extracted from Karhunen-Loève (KL) transform coefficients obtained from co-occurrence matrix of image. Then radial basis neural network (RBNN) trained using these features to discriminate whether the image contains hidden information or not. This system can be used to prevent the suspicious secret communication.

show abstract

Section: Karhunen-loève Transformmentioning

confidence: 99%

Section: Kl-transformmentioning

confidence: 99%

See 1 more Smart Citation

Steganalysis Using KL Transform and Radial Basis Neural Network

Hasoon¹,

Khalifa²

2012

AL-Rafidain Journal of Computer Sciences and Mathematics

View full text Add to dashboard Cite

show abstract

“…On the other hand, image quality for low dose CT is degraded due to noise and noise reduction technique is also an important aspect to improve reconstruction image quality for low dose CT [184]. The CT noise model is setup by Whiting [185] et al and Elbakri [81] et al The scatter noise correction method with blocker array is developed by Zhu et al [186] to estimate scatter noise properties and reduce scatter noise.…”

Section: Recommendations For Further Researchmentioning

confidence: 99%

Geometrically optimized reconstruction algorithms for helical cone beam CT

Yan¹

View full text Add to dashboard Cite

List of Tables 3.1 The maximum pitch values for TPFR and ETPFR algorithms compared with half-scan FDK algorithm and full scan FDK algorithm. . 63 3.2 Comparison of amount of filtering operations for 10 images between FDK, TPFR and ETPFR 65 3.3 The mean value and deviation of full-width half maxima (FWHM) for the TPFR and ETFR algorithm compared with FDK algorithm and tangential filtering algorithm 66 3.4 Parameters for resolution phantom (mm) 73 3.5 Parameters for small resolution phantom 73 3.6 Mean values of the standard derivations of regions reconstructed by FDK, Tangential filtering, TPFR and ETPFR, Zou-Pan (HU) .... 74 4.1 The mean value and deviation of full-width half maxima (FWHM) GTASSR, GTFDK, GTTF and GOFTR algorithms Ill 4.2 Mean values of the standard derivations of regions reconstructed by GTFDK, GTASSR, GTTF and GOFTR(HU) 113 NANYANG TECHNOLOGICAL UNIVERSITY SINGAPORE ATTENTION: The Singapore Copyright Act applies to the use of this document. Nanyang Technological University Library XI List of Figures 1.1 Illustration of Step and shot scanning protocol for single slice CT.. . 4 1.2 Illustration of the helical single slice CT 5 1.3 Illustration of multislice CT 1.4 Comparison collimator thickness between MSCT and single slice CT. 2.1 Illustration of the liner integration and parallel geometry for CT.. . 2.2 Illustration of equalangular fan beam geometry for CT 2.3 Illustration of converting parallel projections from fan beam projections. 2.4 Illustration of fan beam sinogram for half scan 2.5 The geometry of the FDK algorithm 3.1 3D geometry of multislice CT scanner 3.2 In plane geometry of multislice CT scanner 3.3 Illustration the definition of slice thickness S 3.4 Geometric descriptions of u, (x c ,y c , z c) T and (x p ,y p , z p) T 3.5 Tilted construction plane 42 NANYANG TECHNOLOGICAL UNIVERSITY SINGAPORE ATTENTION: The Singapore Copyright Act applies to the use of this document. Nanyang Technological University Library LIST OF FIGURES xn 3.6 Plots of d se (P, 7J) with the dotted curve for 7; = 0.4° and solid curve for 7 , = 0° 45 3.7 Illustration of the in plane geometry for s-t coordinate system. Point F is the projection of the point (£ p , rj pj £ p) on £-r\ plane 50 3.8 Geometry of the C position of the ray at iso-center 52 3.9 In plane geometry of the reformation of the cone parallel projection data set 58 3.10 Center and off-center SSPs for FDK algorithm, tangential filtering algorithm, TPFR algorithm and ETPFR algorithm with p = 1.0, M=64 and S=0.625mm 67 3.11 Images for resolution phantom with z = 0, M = 64 and p = 1.0. (300-900HU), reconstructed by FDK algorithm, Tangential filtering algorithm, TPFR algorithm and ETPFR algorithm, respectively. ... 68 3.12 Images for resolution phantom with z = 0, M = 64 and p = 1.2. (300-900HU), reconstructed by FDK algorithm, Tangential filtering algorithm, TPFR algorithm and ETPFR algorithm, respectively. ... 69 3.13 Images for small resolution phantom with z = 0, M = 64 andp = 1.0. (300-900HU), reconstructed by FDK algorithm, Tangential filterin...

show abstract

A general framework and review of scatter correction methods in x‐ray cone‐beam computerized tomography. Part 1: Scatter compensation approaches

2011

View full text Add to dashboard Cite

Since scattered radiation in cone-beam volume CT implies severe degradation of CT images by quantification errors, artifacts, and noise increase, scatter suppression is one of the main issues related to image quality in CBCT imaging. The aim of this review is to structurize the variety of scatter suppression methods, to analyze the common structure, and to develop a general framework for scatter correction procedures. In general, scatter suppression combines hardware techniques of scatter rejection and software methods of scatter correction. The authors emphasize that scatter correction procedures consist of the main components scatter estimation (by measurement or mathematical modeling) and scatter compensation (deterministic or statistical methods). The framework comprises most scatter correction approaches and its validity also goes beyond transmission CT. Before the advent of cone-beam CT, a lot of papers on scatter correction approaches in x-ray radiography, mammography, emission tomography, and in Megavolt CT had been published. The opportunity to avail from research in those other fields of medical imaging has not yet been sufficiently exploited. Therefore additional references are included when ever it seems pertinent. Scatter estimation and scatter compensation are typically intertwined in iterative procedures. It makes sense to recognize iterative approaches in the light of the concept of self-consistency. The importance of incorporating scatter compensation approaches into a statistical framework for noise minimization has to be underscored. Signal and noise propagation analysis is presented. A main result is the preservation of differential-signal-to-noise-ratio (dSNR) in CT projection data by ideal scatter correction. The objective of scatter compensation methods is the restoration of quantitative accuracy and a balance between low-contrast restoration and noise reduction. In a synopsis section, the different deterministic and statistical methods are discussed with respect to their properties and applications. The current paper is focused on scatter compensation algorithms. The multitude of scatter estimation models will be dealt with in a separate paper.

show abstract

Noise reduction for low-dose helical CT by 3D penalized weighted least-squares sinogram smoothing

Cited by 3 publications

References 18 publications

Steganalysis Using KL Transform and Radial Basis Neural Network

Steganalysis Using KL Transform and Radial Basis Neural Network

Geometrically optimized reconstruction algorithms for helical cone beam CT

A general framework and review of scatter correction methods in x‐ray cone‐beam computerized tomography. Part 1: Scatter compensation approaches

Contact Info

Product

Resources

About